Name - Hettihewa.D.P.G

Index - 190231R

1)

import numpy as np
import cv2 as cv
import matplotlib.pyplot as plt

img = cv.imread(r'emma_gray.jpg')
assert img is not None

t1 = np.linspace(0, 50, 50)
t2 = np.linspace(100, 255, 100)
t3 = np.linspace(150, 255, 106)
 
t = np.concatenate((t1, t2, t3), axis = 0).astype(np.uint8)

fig, ax = plt.subplots(1, 1, figsize = (8,8))
ax.plot(t)
ax.set_aspect('equal')
transImg = cv.LUT(img, t)

fig, ax = plt.subplots(1, 2, figsize = (8,8))
ax[0].imshow(cv.cvtColor(img, cv.COLOR_BGR2RGB))
ax[1].imshow(cv.cvtColor(transImg, cv.COLOR_BGR2RGB))

<matplotlib.image.AxesImage at 0x1ea0eac0e80>

2)

import numpy as np
import cv2 as cv
import matplotlib.pyplot as plt
img = cv.imread(r'brain_proton_density_slice.png')
assert img is not None

hist_f = cv. calcHist([img], [0], None, [256], [0,256])
t1 = np.linspace(0, 0, 140)
t2 = np.linspace(200, 255, 40)
t3 = np.linspace(0, 0, 76)
T1 = np.linspace(0, 0, 180)
T2 = np.linspace(200, 255, 40)
T3 = np.linspace(0, 0, 36)
t = np.concatenate((t1, t2, t3), axis = 0).astype(np.uint8)
T = np.concatenate((T1, T2, T3), axis = 0).astype(np.uint8)
white = cv.LUT(img, t)
gray = cv.LUT(img, T)
fig, ax = plt.subplots(1,3,figsize = (16,8))
ax[0].plot(hist_f)
ax[1].plot(t)
ax[2].plot(T)
fig, ax = plt.subplots(1, 3, figsize = (16,8))
ax[0].imshow(cv.cvtColor(img, cv.COLOR_BGR2RGB))
ax[1].imshow(cv.cvtColor(white, cv.COLOR_BGR2RGB))
ax[2].imshow(cv.cvtColor(gray, cv.COLOR_BGR2RGB))

<matplotlib.image.AxesImage at 0x22df6a6de20>

3)

import numpy as np
import cv2 as cv
import matplotlib.pyplot as plt

img1 = cv.imread(r'highlights_and_shadows.jpg')
assert img1 is not None
gamma = 0.8
limg = cv.cvtColor(img1,  cv.COLOR_BGR2LAB)
L, A, B = cv.split(limg)
t = np.array([(i/255)**gamma*255 for i in np.arange(0,256)]).astype(np.uint8)
g = cv.LUT(L, t)
img = cv.merge([g,A,B])
img = cv.cvtColor(img,  cv.COLOR_LAB2BGR)
hist_f = cv. calcHist([img], [0], None, [256], [0,256])
hist_g = cv. calcHist([g], [0], None, [256], [0,256])
fig, ax = plt.subplots(1,2, figsize = (16, 8))
ax[0].imshow(cv.cvtColor(img1, cv.COLOR_BGR2RGB))
ax[0].set_title('original img')
ax[1].imshow(cv.cvtColor(img, cv.COLOR_BGR2RGB))
ax[0].set_title('original')
ax[1].set_title('gamma corrected')
fig, ax = plt.subplots(1,2,figsize = (16,8))
ax[0].plot(hist_f)
ax[1].plot(hist_g)

[<matplotlib.lines.Line2D at 0x22df805a4f0>]

4)

import cv2 as cv
import numpy as np
import matplotlib.pyplot as plt

im1 = cv.imread(r'shells.png', cv.IMREAD_GRAYSCALE)
assert im1 is not None
hist_f = cv. calcHist([im1], [0], None, [256], [0,256])
for i in range(256):
    if hist_f[i] != 0:
        l = i
        break
for i in range(255,0,-1):
    if hist_f[i] != 0:
        u = i
        break

hist = np.zeros(256)
eq = np.zeros(256)
for i in range(l, u):
    hist[int(round(255*(i-l)/(u-l)))] = hist_f[i]
    eq[i] = 255*(i-l)/(u-l)
im2 = cv.LUT(im1, eq)
im2 = im2.astype(np.uint8)
fig, ax = plt.subplots(1,2,figsize = (16,8))

ax[0].plot(hist_f)
ax[1].plot(hist)
ax[0].set_title('original image histogram')
ax[1].set_title('equalised image histogram')
fig, ax = plt.subplots(1,2,figsize = (16,8))
ax[0].imshow(cv.cvtColor(im1, cv.COLOR_BGR2RGB))
ax[1].imshow(cv.cvtColor(im2, cv.COLOR_BGR2RGB))
ax[0].set_title('original image')
ax[1].set_title('equalised image')

Text(0.5, 1.0, 'equalised image')

5)

import numpy as np
import cv2 as cv
import matplotlib.pyplot as plt

def SSD(im1,im2):
    val = (np.sum((im1.astype("float")-im2.astype("float"))**2))/float(im1.shape[0]*im1.shape[1])
    return val

def zoom_nn(im1, im2, s):
    row = im1.shape[0]*s
    col = im1.shape[1]*s
    imz = np.zeros((row, col, 3), dtype = np.uint8)
    for i in range(row):
        for j in range(col):
            imz[i, j] = im1[int(i/s), int(j/s)]
    fig, ax = plt.subplots(1,3,figsize = (16,8))
    ax[0].imshow(cv.cvtColor(im1[0:row//8, 0:col//8], cv.COLOR_BGR2RGB))
    ax[1].imshow(cv.cvtColor(imz[0:row//2, 0:col//2], cv.COLOR_BGR2RGB))
    ax[2].imshow(cv.cvtColor(im2[0:row//2, 0:col//2], cv.COLOR_BGR2RGB))
    ax[0].axis("off")
    ax[1].axis("off")
    ax[2].axis("off")
    ax[0].set_title('croped original')
    ax[1].set_title('zoomed original')
    ax[2].set_title('croped Large')
    return(imz)
    
im1 = cv.imread(r'a1q5images/im01small.png')
im1l = cv.imread(r'a1q5images/im01.png')
assert im1 is not None
im1z = zoom_nn(im1, im1l, 4)
print("ssd =", SSD(im1z, im1l))

im2 = cv.imread(r'a1q5images/im02small.png')
im2l = cv.imread(r'a1q5images/im02.png')
assert im2 is not None
im2z = zoom_nn(im2, im2l, 4)
print("ssd =", SSD(im2z, im2l))

im3 = cv.imread(r'a1q5images/im03small.png')
im3l = cv.imread(r'a1q5images/im03.png')
assert im2 is not None
im3z = zoom_nn(im3, im3l, 4)

ssd = 408.80714699074076
ssd = 79.33826215277777

import math

def zoom_bl(im1, s):
    row = im1.shape[0]*s
    col = im1.shape[1]*s
    print(row, col)
    imz = np.zeros((row, col, 3), dtype = np.uint8)
    for i in range(row):
        for j in range(col):
            r0, r1 = math.floor(i/s), math.ceil(i/s) 
            fr = int(i/s) - i/s
            c0, c1 = math.floor(j/s), math.ceil(j/s)
            fc = int(j/s) - j/s
            imz[i,j] = (im1[r0, c0]+im1[r1-1, c1-1])
    fig, ax = plt.subplots(1,3,figsize = (16,8))
    ax[0].imshow(cv.cvtColor(imz[0:row//2, 0:col//2], cv.COLOR_BGR2RGB))
im11 = zoom_bl(im3, 4)

6)

import cv2 as cv
import numpy as np
import matplotlib.pyplot as plt

im1 = cv.imread(r'einstein.png', cv.IMREAD_GRAYSCALE).astype(np.float32)
assert im1 is not None

sobel_v = np.array([[-1, -2, -1], [0, 0, 0], [1, 2, 1]], dtype=np.float32)
im_x = cv.filter2D(im1, -1, sobel_v)
sobel_h = np.array([[-1, 0, 1], [-2, 0, 2], [-1, 0, 1]], dtype=np.float32)
im_y = cv.filter2D(im1, -1, sobel_h)
grad = np.sqrt(im_x**2 + im_y**2)
fig, ax = plt.subplots(1,2,figsize = (16,8))
ax[0].imshow(im1, cmap = 'gray', vmin = 0, vmax = 255)
ax[1].imshow(grad, cmap = 'gray')
ax[0].set_title('original image')
ax[1].set_title('filtered image')
R,C = im1.shape[0], im1.shape[1]
im3 = np.zeros((R, C), dtype = im1.dtype)
'''for i in range(R):
    for j in range(C):'''
            

'for i in range(R):\n    for j in range(C):'

R,C = im1.shape[0], im1.shape[1]
im2 = np.zeros((R, C), np.uint8)
for i in range(1, R-1):
    for j in range(1, C-1):
        im_x = im1[i-1, j-1]*(-1) + im1[i-1, j]*(-2) + im1[i-1, j+1]*(-1) + im1[i+1, j-1]*1 + im1[i+1, j]*2 + im1[i+1, j+1]*1
        im_y = im1[i-1, j-1]*(-1) + im1[i, j-1]*(-2) + im1[i-1, j+1]*1 + im1[i+1, j-1]*(-1) + im1[i, j+1]*2 + im1[i+1, j+1]*1
        grad = np.sqrt(im_x**2 + im_x**2)
        im2[i, j] = (grad / 1020) * 255
        
fig, ax = plt.subplots(1, 2, figsize = (16, 8))
ax[0].imshow(im1, cmap = 'gray', vmin = 0, vmax = 255)
ax[1].imshow(im2, cmap = 'gray')
ax[0].set_title("Original Image")
ax[1].set_title("Filtered Image")

Text(0.5, 1.0, 'Sobel Filtered Image')

im3 = np.zeros((R, C), np.uint8)
mat = np.multiply((np.array([[1],[2],[1]], dtype=np.int32)),(np.array([1,0,-1], dtype=np.int32)))
sobel_h1 = mat*-1
sobel_v1 = sobel_h1.T
for i in range(1, R-1):
    for j in range(1,C-1):
        window = np.array([[im1[i-1, j-1], im1[i, j], im1[i+1, j+1]], [im1[i, j-1], im1[i, j], im1[i, j+1]], [im1[i+1, j-1], im1[i+1 ,j], im1[i+1, j+1]]], dtype=np.float32)
        sum_h = np.multiply(window, sobel_h1) 
        sum_v = np.multiply(window, sobel_v1) 
        sum_x, sum_y = 0, 0
        for m in range(3):
            for n in range(3):
                sum_x += sum_h[m, n]
                sum_y += sum_v[m, n] 
        grad = np.sqrt(sum_x**2 + sum_y**2)
        im3[i, j] = (grad / 1020) * 255

fig, ax = plt.subplots(1, 2, figsize = (16, 8))
ax[0].imshow(im1, cmap = 'gray', vmin = 0, vmax = 255)
ax[1].imshow(im3, cmap = 'gray')
ax[0].set_title("Original Image")
ax[1].set_title("Filtered Image")

Text(0.5, 1.0, 'Sobel Filtered Image')

7)

import cv2 as cv
import numpy as np
import matplotlib.pyplot as plt

img = cv.imread(r'daisy.jpg')
assert im1 is not None
mask = np.zeros(img.shape[:2], np.uint8)
rect = (0, 100, 561, 500)

fgdModel = np.zeros((1, 65), np.float64)
bgdModel = np.zeros((1, 65), np.float64)
cv.grabCut(img, mask, rect, bgdModel, fgdModel, 5, cv.GC_INIT_WITH_RECT)

mask1 = np.where((mask==0) | (mask==2), 0, 1).astype("uint8")
img_fgd = img*mask1[:, :, np.newaxis]
mask2 = np.where((mask==1) | (mask==3), 0, 1).astype("uint8")
img_bgd = img*mask2[:, :, np.newaxis]
fig, ax = plt.subplots(1, 4, figsize=(18, 6))

ax[0].imshow(cv.cvtColor(img, cv.COLOR_BGR2RGB))
ax[1].imshow(mask1, cmap = "gray")
ax[2].imshow(cv.cvtColor(img_fgd, cv.COLOR_BGR2RGB))
ax[3].imshow(cv.cvtColor(img_bgd, cv.COLOR_BGR2RGB))
ax[0].set_title("Original")
ax[1].set_title("Mask")
ax[2].set_title("Fore Ground")
ax[3].set_title("Back Ground")

Text(0.5, 1.0, 'Back Ground')

img_blurred = cv.blur(img_bgd, (9, 9), 2)
img_enhanced = cv.add(img_blurred, img_fgd)

fig, ax = plt.subplots(1, 2, figsize=(9, 6))
ax[0].imshow(cv.cvtColor(img, cv.COLOR_BGR2RGB))
ax[1].imshow(cv.cvtColor(img_enhanced, cv.COLOR_BGR2RGB))
ax[0].set_title("Original")
ax[1].set_title("Enhanced")

Text(0.5, 1.0, 'Enhanced')

in this enhacement for background bluring we are using a kernel of 9x9 and when we are convolving image with kernel the darker pixels becomes more darker. There fore we can see that the background just beyond the edges are quiet dark.